LCM (Lightweight Communications and Marshalling) is a lightweight publish/subscribe messaging library created at MIT CSAIL in 2006 for the MIT Talos autonomous vehicle competing in the DARPA Urban Challenge. All messages are sent via UDP multicast — no broker, no central master process. Each message is described in a simple .lcm file (analogous to Protobuf's .proto) and compiled to C, C++, Java, Python, MATLAB, C# or Lua.
LCM's architecture is intentionally minimal: zero ROS dependencies, no name server, no dynamic node discovery beyond standard IP multicast. This makes LCM an ideal choice for embedded systems where a full DDS or ROS 2 stack is too heavy. LCM is widely used as the communication layer in proprietary robotics company stacks (Boston Dynamics historically, some internal Toyota Research Institute tools), in humanoid robotics (Atlas DARPA Robotics Challenge), in autonomous vehicle research (Mercedes-Benz Bertha Drive Project), and in academia as a lightweight alternative to ROS 1.
The wire format is deterministic and versioned — there is an lcm-spy tool for traffic inspection and lcm-logger/lcm-logplayer for record/replay (analogous to rosbag). LCM has no built-in QoS, security or discovery mechanism — a deliberate design decision that simplifies implementation at the cost of limited use in geographically distributed networks. In typical robotic applications (single host or several nodes in a single subnet) LCM is one of the fastest transport layers available.
The library is open source under LGPL 2.1, maintained by the community on GitHub (lcm-proj/lcm) with contributions from MIT, Boston Dynamics, TRI and Drake Project engineers. The latest release 1.5.1 (February 2024) added support for newer Python and Java distributions. LCM is a required dependency of the Drake project (Toyota Research / MIT), where it is used for internal communication between planning and control modules.
In the ROS ecosystem LCM has several community wrappers (ros1_bridge, lcm_to_ros, ros2_lcm_bridge), but is not an official ROS middleware — it functions as a parallel, independent technology. In practice, robotics teams often mix stacks — they use ROS 2 for the application layer and LCM for fast low-level communication between real-time control modules.
Middleware is a software layer that mediates between applications, services, sensors, drivers, and execution layers. In robotics, middleware is typically responsible for inter-process communication, message passing, hardware abstraction, and module integration within a single system. The most widely used robotics middleware is ROS (Robot Operating System), which provides a publish-subscribe message bus, service calls, and a rich ecosystem of packages.
A Transport Layer is the software layer responsible for data transmission between elements of a robotic or software system — such as processes, nodes, devices, and services. It handles serialisation, routing, reliability, and latency guarantees for messages. In robotics examples include DDS (Data Distribution Service) used by ROS 2, LCM (Lightweight Communications and Marshalling), ZeroMQ, and proprietary message buses on embedded platforms.
An API Library is a software package that exposes programmatic interfaces for communicating with a device, service, or system. In robotics it typically forms a lightweight integration layer built on top of the manufacturer's official API or an open-source project, abstracting low-level protocol details and providing language-native bindings (Python, C++, Java, etc.).
Device Integration denotes the role of software responsible for communication, configuration, initialization and handling of specific devices, sensors, controllers or hardware components within a robotic system.
Family of open communication protocols and middleware libraries not tied to a specific ecosystem (LCM, ZeroMQ, MQTT, Zenoh-pico).
MIT Talos (DARPA Urban Challenge 2007) — first public use of LCM as the communication layer for an autonomous vehicle. Boston Dynamics Atlas (DARPA Robotics Challenge 2013-2015) — LCM used as the internal middleware for messaging between control, perception and planning modules on the Atlas humanoid. Drake (Toyota Research Institute / MIT) — required dependency. Drake uses LCM as the primary communication mechanism between simulation, planning and control modules. Mercedes-Benz Bertha Drive Project (2013) — autonomous driving along the historic Bertha Benz route, LCM as the communication layer in the prototype. Stanford Junior (DARPA Urban Challenge) — the second autonomous vehicle using LCM (after MIT Talos). Numerous academic projects in humanoid robotics and autonomous vehicles — DRC Tartan Rescue (CMU), Florida Institute for Human and Machine Cognition (IHMC).
GitHub: lcm-proj/lcm — ~1.8k stars, ~600 forks, ~80 contributors, active issues mostly around bindings for new language versions. Academic community — several thousand citations in robotics and autonomous driving papers; the original paper "LCM: Lightweight Communications and Marshalling" (Huang, Olson, Moore, 2010, IROS) is cited over 600 times. Professional ecosystem — Drake (TRI/MIT), historically Boston Dynamics, IHMC, several AV startups. Google Groups mailing list (lcm-users) — low activity, occasional questions. Maintenance — team of ~5-8 active contributors (mostly TRI and Drake project engineers), release cadence ~1 release / year since 2017.
Boston Dynamics bipedal humanoid robot. The fully electric generation unveiled in 2024 succeeds the hydraulic Atlas that was retired after more than a decade of research.
Tutor Intelligence industrial robot designed for palletizing, depalletizing and case-picking in warehouses and factories, offered as Robotics-as-a-Service from $14/hr.
Ubuntu 24.04 LTS 'Noble Numbat' — supported until April 2029. The host for ROS 2 Jazzy.
Requires a network with UDP IPv4 multicast enabled on the loopback (lo) and/or Ethernet interface. Some Wi-Fi routers filter multicast — TCP fallback mode is required in those cases.
License family: Weak Copyleft
Python 3.12 support, C# and Lua generator fixes, minor CMake improvements.
Full Apple Silicon (arm64) support, documentation updates, new experimental lcm-gen-rust.
Core refactor for C++17, improved Windows support, multicast discovery bug fixes.
Debian/Ubuntu packaging, lcm-spy GUI improvements, Lua generator, Java 8 baseline.
First stable public release on GitHub. C, C++, Java, Python bindings + lcm-gen.
First internal MIT version — used in DARPA Urban Challenge by team Talos.
